Looping over columns and changing Null values - sql

Question

I have a table called myTable where some values are null.
I want to replace all null values in all columns with the previous non null value. I found some code that iterates over each row for a specific column, and changes Null Values as I want.
DECLARE #value AS int
UPDATE myTable
SET
#value = COALESCE(col2, #value),
col2 = COALESCE(col2, #value)
Result:
This does what I want it do do but only for one column at the time. My goal is to alter the code above in some way so that I can automatically loop over each column in the table.
I ran into several issues when trying to achieve this. Here is my attempt
DECLARE #ColNames table (NAMES nvarchar(50), ARRAYINDEX int identity(1,1) )
INSERT INTO #ColNames (NAMES)
VALUES ('col1'),('col2'),('col3')
DECLARE #INDEXVAR int
DECLARE #TOTALCOUNT int
SET #INDEXVAR = 0
SELECT #TOTALCOUNT = COUNT(*) FROM #ColNames
WHILE #INDEXVAR < #TOTALCOUNT
BEGIN
DECLARE #curColName nvarchar (50)
SELECT #INDEXVAR = #INDEXVAR + 1
SELECT #curColName = NAMES from #ColNames where ARRAYINDEX = #INDEXVAR
DECLARE #value AS int
UPDATE myTable
SET
#value = COALESCE(#curColName, #value),
#curColName = COALESCE(#curColName, #value)
END
The issues that I have found and not been able to solve are the following:
#curColName is just a nvarchar variable and not a representation of my actual column, even if the names are the same. This gives me errors on both lines inside the SET statement.
When hard coding the column names in my loop inside the BEGIN/END statement., the script fills out ALL Null values with a number. So col2 gets the value 3 on ALL rows, not only row 2 and 3 as my previous example.
If these two points are hard or impossible to solve, is there an easier way of solving this problem?
Thanks

Answer 1

This is based on your expected results, and that you actually want to assign the value of Col2 to be the value of the previous non-NULL value, when ordered by the column pk.
If so, to achieve this you can use an updatable CTE. The first CTE puts the data into groups, based on the non-NULL values, and then the second gets the MAX value of Col2 in the group (which would be the non-NULL value). Finally you UPDATE against that CTE on rows where Col2 has the value NULL:
CREATE TABLE dbo.YourTable (PK int,
Col1 int,
Col2 int);
INSERT INTO dbo.YourTable (PK,Col1,Col2)
VALUES(1,2,NULL),
(2,NULL,3),
(3,NULL,NULL);
GO
WITH Groups AS(
SELECT Col2,
COUNT(Col2) OVER (ORDER BY PK) AS Grp
FROM dbo.YourTable),
Maxes AS(
SELECT Col2,
MAX(Col2) OVER (PARTITION BY Grp) AS MaxCol2
FROM Groups)
UPDATE Maxes
SET Col2 = MaxCol2
WHERE Col2 IS NULL;
GO
SELECT *
FROM dbo.YourTable;
GO
DROP TABLE dbo.YourTable;